Data Analyst interview questions

Data analyst interviews test three things: can you get the right data (SQL), can you reason about it (statistics and business sense), and can you explain it to people who are not analysts. Expect a SQL exercise and at least one case question.

  1. 1What is the difference between INNER JOIN, LEFT JOIN and FULL OUTER JOIN?

    What a good answer shows: Which rows each keeps, a quick example, and the classic mistake of filtering a LEFT JOIN in the WHERE clause.

  2. 2How would you find the second highest salary in a table?

    What a good answer shows: A window function (DENSE_RANK) or a subquery, and how ties and NULLs are handled.

  3. 3Explain GROUP BY and HAVING.

    What a good answer shows: GROUP BY aggregates rows; HAVING filters the groups after aggregation, WHERE filters rows before.

  4. 4What are window functions, and when do you use them?

    What a good answer shows: Running totals, rankings and comparisons with the previous row (LAG/LEAD) without collapsing rows.

  5. 5How do you deal with missing or messy data?

    What a good answer shows: Find out why it is missing, then decide: drop, fill, flag or fix at the source. Explain how each choice changes the result.

  6. 6What is the difference between mean, median and mode? When is the median better?

    What a good answer shows: The median resists outliers, so it suits skewed data like salaries or order values.

  7. 7Explain correlation versus causation with an example.

    What a good answer shows: Two metrics moving together is not proof that one causes the other; mention confounders and experiments.

  8. 8How would you design an A/B test?

    What a good answer shows: Hypothesis, one primary metric, sample size, random assignment, run time and how you would read the result.

  9. 9What is a p-value, in plain words?

    What a good answer shows: The chance of seeing a result at least this extreme if there were really no effect, and why a small p-value is not the same as a big effect.

  10. 10Sales dropped 10% last week. How would you investigate?

    What a good answer shows: Check the data first, then split by time, region, channel, product and customer type until the drop is explained.

  11. 11How do you choose the right chart?

    What a good answer shows: Lines for trends, bars for comparisons, scatter for relationships, and avoiding pie charts with many slices.

  12. 12Tell me about a dashboard you built. Who used it and what changed?

    What a good answer shows: The audience, the few metrics that mattered, and a decision someone made because of it.

  13. 13How do you define a good metric?

    What a good answer shows: Tied to a goal, easy to understand, hard to game, and measured the same way over time.

  14. 14How would you explain a complex finding to a non-technical manager?

    What a good answer shows: Start with the answer and what to do, one chart, then the detail only if they ask.

  15. 15What tools do you use, and why?

    What a good answer shows: SQL plus a BI tool (Power BI, Tableau, Looker) and Excel or Python; show you pick tools for the task.

  16. 16How do you check your analysis is correct?

    What a good answer shows: Reconcile totals with a known source, check row counts after joins, sanity-check outliers and ask someone to review.

  17. 17What is the difference between a data warehouse and a database?

    What a good answer shows: Transactional databases run the app; warehouses store history in a shape built for analysis.

  18. 18How would you measure the success of a new feature?

    What a good answer shows: Adoption, the behaviour it should change, a guardrail metric, and a comparison group.

  19. 19Tell me about a time your analysis changed a decision.

    What a good answer shows: Use STAR and include the number: "We moved budget to the channel with 2x the return".

  20. 20What would you want to analyse first if you joined us?

    What a good answer shows: Show you looked at the business: a sensible question tied to their growth or customers.